Skip to content

Additional Research Services

The NCShare Compute Cluster is a general purpose high-throughput installation and is container ready to host workloads for a broad array of scientific projects.

Quick facts:

  • 8 compute nodes with 128 cores and 512 GB RAM each (1024 cores and 4 TB of RAM total)
  • Currently in testing phase: 4 H200 w/141GB of NVRAM GPU nodes with 8 H200 GPUS and 2TB RAM each (32 H200 GPUs total)
  • Interconnects are 10 Gbps or 40 Gbps
  • Utilizes a 400 TB FreeNAS NFS share
  • Systems run Ubuntu and SLURM is the job scheduler
  • GPU: Driver Version: 570.133.20 CUDA Version: 12.8

The NCShare Compute Cluster is managed and supported by a regional consortium. We are currently in an early adopters phase of deployment and actively seeking researchers from participating institutions to try out the resources.

While we are in this phase, substantial grants for resource hours are available.

Review are guides for Cluster Computing